E8500 Wolfdale Preview... WOW!!!

[dominick32]

I am sure most of you have seen my QX9650 on Water thread and have seen the massive capability of Penryn/Yorkfields 45nm halfnium technology. Even on my quad core I was able to achieve astounding clocks. In a short wrap up of my QX experience:

4.0 GHz 24/7 stability is available to you @ 1.36 vcore (stock max intel recommended volts)

4.2 GHz 24/7 stability is available @ 1.48 vcore (stable but 66*C to 70*C)

4.56 GHz Full gaming (ie: crysis), benching, 3d stability for hours @ 1.60 vcore

4.70 GHz SuperPi 1m 9 Second Barrier Broken in Vista (maximum SuperPi1m shot)

4.79 GHz Maximum clockspeed capable under water cooling for screenshot.

As you can see, even with the higher TDP of the quad cores they are extremely efficient in the new 45nm process.

All of the following information was taken from member Chrisch @ XtremeSystems.org in his results thread. The guy got his hands on one of the first E8500 Wolfdale ES chips. He ran on it a P5KE and water cooling. Here comes the madness:

SUPERPI 32M @ 4.75 GHz,,, wow..


Their you have it folks, assuming retail wolfdales will perform relatively the same as this ES counterpart you should assume to expect around 200 to 300 mhz more than a quad on your overclock with the lower TDP dual core. As you can see, he also had no problem taking FSB all the way up to 517 on the dual core where as my quad was limited to 450/455. So, I think that throws the FSB wall question at under 5.0 Ghz right out of the water. We still dont have any LN2, DI, Phase data so hopefully it still does not have a wall at higher FSB's.

 

[dominick32]

Oh I forgot to mention.
How does 4.1 GHz full 24/7 stability at stock vcore sound?
and
How does 4.2 GHz full 24/7 stability sound at 1.40 vcore?? A full .10 volts below my quad!!!
and
I am also assuming these E8500's will bench 3dmark @ 4.70 GHz +.....

He has approximately 200 mhz more firepower at 1.40 vcore than my QX.

I cant wait until these things hit the retail shelf.

 

[bing]

Thanks Dom, Intel 45nm really delivers a huge punch in cpu perfomance !

Guys, I feel its time to start stalking the classified for cheap 45nm quad offer !

 

[ratbuddy]

How did he get a 9.5 multiplier?

 

[dominick32]

How did he get a 9.5 multiplier?

Intel is reverting back to 1/2 multipliers.

E8500 = Stock 9.5 Multi X 333 FSB = 3.16 GHz
It is also the highest binned consumer E wolfdale available. So, think of it as a very cheap dual core Extreme Processor without the unlocked multi.

 

[Kuroimaho]

Holy moly, 1-2 more revision or stepping and 5G capable duals ?

 

[Jhonas]

No chance in fighting this sort of power...
I wants one... not getting a mortage out for one kinda helps, unlike a top end quad.
Very nicely done Dom. Kudos +2...

 

[majinwar]

4.9Ghz!

 

[trickson]

That I amazing! Don't think AMD can compete with that at all .
Where can I get one of them ?

 

[jordon]

i would do terrible, terrible things for this chip

 

[Rattle]

with a chip like this I am starting not to care I cant go 45nm quad on this mobo. How much are those going to be ?

 

[trickson]

i would do terrible, terrible things for this chip

OH ME TO!

 

[Super Nade]

I wonder if the lower bins can scale similarly?

 

[xilix]

Not too shabby. I was expecting more on the OC.

 

[dominick32]

Not too shabby. I was expecting more on the OC.

I think once we see the retails hit the market, we will definitely start to see 4.3 to 4.5 GHz 24/7 stable duallys on air/water. Compared to the 4.0 to 4.25 GHz on the QX's. I was pretty much expecting only around 200 to 300 mhz more OC from lower TDP on the duallys, and thats pretty damn close what it looks to be from his results.

I also think his SuperPi 1m is misleading, because he runs a SuperPi32m at basically the same clock speed: 4.75 GHz. That tells me that I think he just never really maxed out Spi1m yet. Taking that into account, I am assuming 4.70 GHz 3dmark benching is possible, and 4.80 to 4.90 GHz OS boot + superpi1m is possible as well as a 5.0 GHz validation shot with SetFSB.

with a chip like this I am starting not to care I cant go 45nm quad on this mobo. How much are those going to be ?

Dont quote me on this rattle, but I read somewhere around $315.

That I amazing! Don't think AMD can compete with that at all .
Where can I get one of them ?

Unfortunately, unless you are buddies with someone that works for an Intel Distributor who has signed an NDA, than you will have to wait until January 2008 for retails to ship.

I wonder if the lower bins can scale similarly?

Hey Nade.
Back from vacation yet? LOL
Forgetting about FSB and multi's for a minute here and concentrating on process. I am assuming the lower E consumer processors will all pretty much guarantee 4.0 GHz to 4.1 GHz no problem at close to stock vcore, but upper level benchmarking (4.5 GHz+) and closer to 4.3 to 4.5GHz prime stability might be harder to come by with the lower grade chips.

Now, if you factor in multipliers its pretty much: How much money are you willing to spend on a dual core Wolfdale to unlock your multiplier 1 notch upward? Because if you think about it, these lower grade dual cores will all probably be able to max out multi/fsb completely. Intel has another winner this time around.

Think about it, pretend you have a motherboard that craps out at 500fsb, here are your theoretical maximums. What would you pay for?
E8500 500 X 9.5 = 4.75 GHz
E8400 500 X 9 = 4.50 GHz
E8300 500 X 8.5 = 4.25 GHz
E8200 500 X 8 = 4.00 GHz

That means, if you buy the cheapest processor E8200 with 8X multi, in order to clock 4.00 GHz 24/7 you would need to be running 500 FSB 24 hours a day, 7 days a week. The 45nm halfnium should be good to 4.4 GHz benching at least on the E8200, so you would need theoretically 550 FSB to run a simple 3dmark bench @ 4.40 GHz... Again, even with the duallys, you are pretty much paying for your multiplier. But, unlike with C2D's in the past this time you can obviously see spending a little extra money for a higher multi will get you some serious performance.

 

[Super Nade]

Holy moly! Good info dom. Ofcourse, you are predicting, but the emphasis is on the process, so as you said I doubt you will see huge variations.

Well, my board should be good upto 500+ FSB, so the only constraint will be my poor pocket.

 

[ALiEN2953]

ES?

 

[dominick32]

ES?

This stands for Engineering Sample. Not sure you if you were asking what ES means. With the QX9650, the retails are actually proving to be better clockers than the ES's, so lets hope the same thing happens with these retail E's.

 

[MadMan007]

Engineering sample, and for all or at least most previous C2D releases they have clocked higher than retail CPUs, even the top-bin ones although maybe not Xtremes. Seems about right, 200-300MHz more stable is about the same difference we see with current G0 dual vs quad. Or it could be that Intel is continuously improving the new process and the retail chips will do closer to the ES than they have in the past.

 

[Valk]

man. results like this make it so hard to decide on a dual or a quad X_X. cant wait to get one though and mess with it under my G5.